Introduction

Since Jan. 1, 2015, The Washington Post has been compiling a database of every fatal shooting in the US by a police officer in the line of duty.

While there are many challenges regarding data collection and reporting, The Washington Post has been tracking more than a dozen details about each killing. This includes the race, age and gender of the deceased, whether the person was armed, and whether the victim was experiencing a mental-health crisis. The Washington Post has gathered this supplemental information from law enforcement websites, local new reports, social media, and by monitoring independent databases such as "Killed by police" and "Fatal Encounters". The Post has also conducted additional reporting in many cases.

There are 4 additional datasets: US census data on poverty rate, high school graduation rate, median household income, and racial demographics. Source of census data.

Upgrade Plotly

Run the cell below if you are working with Google Colab

%pip install --upgrade plotly

Requirement already satisfied: plotly in c:\users\shruti soni\appdata\local\programs\python\python310\lib\site-packages (5.5.0)
Requirement already satisfied: six in c:\users\shruti soni\appdata\local\programs\python\python310\lib\site-packages (from plotly) (1.16.0)
Requirement already satisfied: tenacity>=6.2.0 in c:\users\shruti soni\appdata\local\programs\python\python310\lib\site-packages (from plotly) (8.0.1)
Note: you may need to restart the kernel to use updated packages.

Import Statements

import numpy as np
import pandas as pd
import plotly.express as px
import matplotlib.pyplot as plt
import seaborn as sns

# This might be helpful:
from collections import Counter

Notebook Presentation

pd.options.display.float_format = '{:,.2f}'.format

Load the Data

df_hh_income = pd.read_csv('Median_Household_Income_2015.csv', encoding="windows-1252")
df_pct_poverty = pd.read_csv('Pct_People_Below_Poverty_Level.csv', encoding="windows-1252")
df_pct_completed_hs = pd.read_csv('Pct_Over_25_Completed_High_School.csv', encoding="windows-1252")
df_share_race_city = pd.read_csv('Share_of_Race_By_City.csv', encoding="windows-1252")
df_fatalities = pd.read_csv('Deaths_by_Police_US.csv', encoding="windows-1252")

Preliminary Data Exploration

• What is the shape of the DataFrames?
• How many rows and columns do they have?
• What are the column names?
• Are there any NaN values or duplicates?

#income
df_hh_income.shape

(29322, 3)

df_hh_income.columns

Index(['Geographic Area', 'City', 'Median Income'], dtype='object')

df_hh_income.isnull().sum()

Geographic Area     0
City                0
Median Income      51
dtype: int64

df_hh_income.duplicated().sum()

0

df_pct_poverty.shape

(29329, 3)

df_pct_poverty.columns

Index(['Geographic Area', 'City', 'poverty_rate'], dtype='object')

df_pct_poverty.isnull().sum()

Geographic Area    0
City               0
poverty_rate       0
dtype: int64

df_pct_poverty.duplicated().sum()

0

df_pct_completed_hs.shape

(29329, 3)

df_pct_completed_hs.columns

Index(['Geographic Area', 'City', 'percent_completed_hs'], dtype='object')

df_pct_completed_hs.isnull().sum()

Geographic Area         0
City                    0
percent_completed_hs    0
dtype: int64

df_pct_completed_hs.duplicated().sum()

0

df_share_race_city.shape

(29268, 7)

df_share_race_city.columns

Index(['Geographic area', 'City', 'share_white', 'share_black',
       'share_native_american', 'share_asian', 'share_hispanic'],
      dtype='object')

df_share_race_city.isnull().sum()

Geographic area          0
City                     0
share_white              0
share_black              0
share_native_american    0
share_asian              0
share_hispanic           0
dtype: int64

df_share_race_city.duplicated().sum()

0

df_fatalities.shape

(2535, 14)

df_fatalities.columns

Index(['id', 'name', 'date', 'manner_of_death', 'armed', 'age', 'gender',
       'race', 'city', 'state', 'signs_of_mental_illness', 'threat_level',
       'flee', 'body_camera'],
      dtype='object')

df_fatalities.isnull().sum()

id                           0
name                         0
date                         0
manner_of_death              0
armed                        9
age                         77
gender                       0
race                       195
city                         0
state                        0
signs_of_mental_illness      0
threat_level                 0
flee                        65
body_camera                  0
dtype: int64

df_fatalities.duplicated().sum()

0

Data Cleaning - Check for Missing Values and Duplicates

Consider how to deal with the NaN values. Perhaps substituting 0 is appropriate.

df_hh_income = pd.read_csv('Median_Household_Income_2015.csv', encoding="windows-1252")
df_pct_poverty = pd.read_csv('Pct_People_Below_Poverty_Level.csv', encoding="windows-1252")
df_pct_completed_hs = pd.read_csv('Pct_Over_25_Completed_High_School.csv', encoding="windows-1252")
df_share_race_city = pd.read_csv('Share_of_Race_By_City.csv', encoding="windows-1252")
df_fatalities = pd.read_csv('Deaths_by_Police_US.csv', encoding="windows-1252")

df_hh_income.fillna(value=0,inplace=True) 

df_pct_poverty.fillna(value=0,inplace=True) 

df_pct_completed_hs.fillna(value=0,inplace=True) 

df_share_race_city.fillna(value=0,inplace=True) 

df_fatalities .fillna(value=0,inplace=True) 

Chart the Poverty Rate in each US State

Create a bar chart that ranks the poverty rate from highest to lowest by US state. Which state has the highest poverty rate? Which state has the lowest poverty rate? Bar Plot

df_pct_poverty.drop(df_pct_poverty[df_pct_poverty["poverty_rate"]=="-"].index, inplace = True)

df_pct_poverty["poverty_rate"]=pd.to_numeric(df_pct_poverty["poverty_rate"])
df_pct_poverty_fg = df_pct_poverty.groupby('Geographic Area')["poverty_rate"].mean().reset_index()

df_pct_poverty_fg 

	Geographic Area	poverty_rate
0	AK	19.85
1	AL	20.65
2	AR	22.96
3	AZ	25.67
4	CA	17.12
5	CO	13.36
6	CT	9.14
7	DC	18.00
8	DE	12.56
9	FL	17.57
10	GA	23.78
11	HI	13.40
12	IA	12.29
13	ID	18.24
14	IL	13.88
15	IN	15.50
16	KS	14.76
17	KY	20.08
18	LA	22.34
19	MA	9.59
20	MD	10.31
21	ME	16.89
22	MI	17.90
23	MN	13.75
24	MO	20.11
25	MS	26.88
26	MT	16.51
27	NC	19.75
28	ND	12.16
29	NE	12.98
30	NH	12.66
31	NJ	8.19
32	NM	23.08
33	NV	12.47
34	NY	11.67
35	OH	14.85
36	OK	20.66
37	OR	16.52
38	PA	12.52
39	RI	10.37
40	SC	22.16
41	SD	16.03
42	TN	19.89
43	TX	19.92
44	UT	11.98
45	VA	14.59
46	VT	13.79
47	WA	15.02
48	WI	12.86
49	WV	21.13
50	WY	9.89

plt.figure(figsize=(12, 6))
sns.lineplot(x='Geographic Area', y='poverty_rate', data=df_pct_poverty)
plt.legend(bbox_to_anchor=(1.05, 1), loc='upper left')
plt.xticks(rotation=90)
plt.show()

No artists with labels found to put in legend.  Note that artists whose label start with an underscore are ignored when legend() is called with no argument.

Chart the High School Graduation Rate by US State

Show the High School Graduation Rate in ascending order of US States. Which state has the lowest high school graduation rate? Which state has the highest?

df_pct_completed_hs

	Geographic Area	City	percent_completed_hs
0	AL	Abanda CDP	21.2
1	AL	Abbeville city	69.1
2	AL	Adamsville city	78.9
3	AL	Addison town	81.4
4	AL	Akron town	68.6
...	...	...	...
29324	WY	Woods Landing-Jelm CDP	100
29325	WY	Worland city	85.6
29326	WY	Wright town	89.2
29327	WY	Yoder town	79.4
29328	WY	Y-O Ranch CDP	100

29329 rows × 3 columns

df_pct_completed_hs.drop(df_pct_completed_hs[df_pct_completed_hs["percent_completed_hs"]=="-"].index, inplace = True)
df_pct_completed_hs["pecent_completed_hs"]=pd.to_numeric(df_pct_completed_hs["percent_completed_hs"])
df_pct_completed_hs_fg= df_pct_completed_hs.groupby('Geographic Area')["percent_completed_hs"].mean().reset_index()
#We can observe that the average price has reduced

plt.figure(figsize=(12, 6))
sns.lineplot(x='Geographic Area', y='percent_completed_hs', data=df_pct_completed_hs_fg)
plt.legend(bbox_to_anchor=(1.05, 1), loc='upper left')
plt.xticks(rotation=90)
plt.show()

No artists with labels found to put in legend.  Note that artists whose label start with an underscore are ignored when legend() is called with no argument.

Visualise the Relationship between Poverty Rates and High School Graduation Rates

Create a line chart with two y-axes to show if the rations of poverty and high school graduation move together.

# create figure and axis objects with subplots()
from matplotlib.pyplot import figure
fig,ax = plt.subplots(figsize=(10,6))
plt.figure(figsize=(15, 6), dpi=80)
# make a plot
ax.plot(df_pct_completed_hs_fg['Geographic Area'],
             df_pct_completed_hs_fg['percent_completed_hs'],color="red", marker="o")
# set x-axis label
ax.set_xlabel('States',fontsize=14)
# set y-axis label
ax.set_ylabel("Percent Completion High School",color="red",fontsize=14)

# twin object for two different y-axis on the sample plot
ax2=ax.twinx()
# make a plot with different y-axis using second axis object
ax2.plot(df_pct_poverty_fg['Geographic Area'],df_pct_poverty_fg['poverty_rate'],color="blue",marker="o")
ax2.set_ylabel("Poverty",color="blue",fontsize=14)
ax.tick_params(axis='x', which='major', labelsize=10,rotation=90)

plt.show()

<Figure size 1200x480 with 0 Axes>

Now use a Seaborn .jointplot() with a Kernel Density Estimate (KDE) and/or scatter plot to visualise the same relationship

sns.jointplot(x=df_pct_poverty_fg["poverty_rate"],y=df_pct_completed_hs_fg['percent_completed_hs'], color = 'blue')
plt.show()

  

data=pd.DataFrame([df_pct_poverty_fg["poverty_rate"],df_pct_completed_hs_fg['percent_completed_hs']])
data=data.T
data

	poverty_rate	percent_completed_hs
0	19.85	84.63
1	20.65	80.30
2	22.96	79.95
3	25.67	80.47
4	17.12	81.96
5	13.36	90.11
6	9.14	91.59
7	18.00	89.30
8	12.56	88.52
9	17.57	85.74
10	23.78	79.01
11	13.40	91.67
12	12.29	90.11
13	18.24	85.17
14	13.88	88.48
15	15.50	86.32
16	14.76	88.23
17	20.08	82.37
18	22.34	79.29
19	9.59	92.40
20	10.31	88.42
21	16.89	91.43
22	17.90	89.21
23	13.75	89.47
24	20.11	83.52
25	26.88	78.47
26	16.51	90.49
27	19.75	83.25
28	12.16	87.82
29	12.98	89.99
30	12.66	90.71
31	8.19	90.85
32	23.08	80.98
33	12.47	87.72
34	11.67	90.61
35	14.85	88.34
36	20.66	82.91
37	16.52	88.30
38	12.52	89.02
39	10.37	88.82
40	22.16	80.85
41	16.03	87.75
42	19.89	81.63
43	19.92	75.69
44	11.98	91.62
45	14.59	84.88
46	13.79	89.98
47	15.02	88.20
48	12.86	90.26
49	21.13	82.35
50	9.89	92.10

Seaborn's .lmplot() or .regplot() to show a linear regression between the poverty ratio and the high school graduation ratio.

sns.lmplot(x="poverty_rate",y='percent_completed_hs',data=data)
plt.show()

sns.regplot(x="poverty_rate",y='percent_completed_hs',data=data)
plt.show()

Create a Bar Chart with Subsections Showing the Racial Makeup of Each US State

Visualise the share of the white, black, hispanic, asian and native american population in each US State using a bar chart with sub sections.

df_share_race_city = pd.read_csv('Share_of_Race_By_City.csv', encoding="windows-1252")

df_share_race_city

	Geographic area	City	share_white	share_black	share_native_american	share_asian	share_hispanic
0	AL	Abanda CDP	67.2	30.2	0	0	1.6
1	AL	Abbeville city	54.4	41.4	0.1	1	3.1
2	AL	Adamsville city	52.3	44.9	0.5	0.3	2.3
3	AL	Addison town	99.1	0.1	0	0.1	0.4
4	AL	Akron town	13.2	86.5	0	0	0.3
...	...	...	...	...	...	...	...
29263	WY	Woods Landing-Jelm CDP	95.9	0	0	2.1	0
29264	WY	Worland city	89.9	0.3	1.3	0.6	16.6
29265	WY	Wright town	94.5	0.1	1.4	0.2	6.2
29266	WY	Yoder town	97.4	0	0	0	4
29267	WY	Y-O Ranch CDP	92.8	1.5	2.6	0	11.8

29268 rows × 7 columns

df_share_race_city.drop(df_share_race_city[df_share_race_city["share_white"]=="(X)" ].index, inplace = False)
df_share_race_city["share_white"]=pd.to_numeric(df_share_race_city["share_white"])
df_share_race_city["share_black"]=pd.to_numeric(df_share_race_city["share_black"])
df_share_race_city["share_native_american"]=pd.to_numeric(df_share_race_city["share_native_american"])
df_share_race_city["share_asian"]=pd.to_numeric(df_share_race_city["share_asian"])
df_share_race_city["share_hispanic"]=pd.to_numeric(df_share_race_city["share_hispanic"])
df_share_race_city_fg= df_share_race_city.groupby(['Geographic area'])

df_share_race_city

	Geographic area	City	share_white	share_black	share_native_american	share_asian	share_hispanic
0	AL	Abanda CDP	67.2	30.2	0	0	1.6
1	AL	Abbeville city	54.4	41.4	0.1	1	3.1
2	AL	Adamsville city	52.3	44.9	0.5	0.3	2.3
3	AL	Addison town	99.1	0.1	0	0.1	0.4
4	AL	Akron town	13.2	86.5	0	0	0.3
...	...	...	...	...	...	...	...
29263	WY	Woods Landing-Jelm CDP	95.9	0	0	2.1	0
29264	WY	Worland city	89.9	0.3	1.3	0.6	16.6
29265	WY	Wright town	94.5	0.1	1.4	0.2	6.2
29266	WY	Yoder town	97.4	0	0	0	4
29267	WY	Y-O Ranch CDP	92.8	1.5	2.6	0	11.8

29268 rows × 7 columns

Create Donut Chart by of People Killed by Race

Hint: Use .value_counts()

	id	name	date	manner_of_death	armed	age	gender	race	city	state	signs_of_mental_illness	threat_level	flee	body_camera
0	3	Tim Elliot	02/01/15	shot	gun	53.00	M	A	Shelton	WA	True	attack	Not fleeing	False
1	4	Lewis Lee Lembke	02/01/15	shot	gun	47.00	M	W	Aloha	OR	False	attack	Not fleeing	False
2	5	John Paul Quintero	03/01/15	shot and Tasered	unarmed	23.00	M	H	Wichita	KS	False	other	Not fleeing	False
3	8	Matthew Hoffman	04/01/15	shot	toy weapon	32.00	M	W	San Francisco	CA	True	attack	Not fleeing	False
4	9	Michael Rodriguez	04/01/15	shot	nail gun	39.00	M	H	Evans	CO	False	attack	Not fleeing	False
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
2530	2822	Rodney E. Jacobs	28/07/17	shot	gun	31.00	M	0	Kansas City	MO	False	attack	Not fleeing	False
2531	2813	TK TK	28/07/17	shot	vehicle	0.00	M	0	Albuquerque	NM	False	attack	Car	False
2532	2818	Dennis W. Robinson	29/07/17	shot	gun	48.00	M	0	Melba	ID	False	attack	Car	False
2533	2817	Isaiah Tucker	31/07/17	shot	vehicle	28.00	M	B	Oshkosh	WI	False	attack	Car	True
2534	2815	Dwayne Jeune	31/07/17	shot	knife	32.00	M	B	Brooklyn	NY	True	attack	Not fleeing	False

2535 rows × 14 columns

values=df_fatalities["race"].value_counts()

values=dict(values)

values.values()

dict_values([1201, 618, 423, 195, 39, 31, 28])

import matplotlib.pyplot as plt
  
 
# given values
sizes = values.values()
  
# Setting labels for items in Chart
labels = values.keys()

  
# explosion
explode = (0.05, 0.05, 0.05, 0.05, 0.05)
  
# Pie Chart
plt.pie(sizes, labels=labels,
        autopct='%1.1f%%', pctdistance=0.85)
  
# draw circle
centre_circle = plt.Circle((0, 0), 0.70, fc='white')
fig = plt.gcf()
  
# Adding Circle in Pie chart
fig.gca().add_artist(centre_circle)
  
# Adding Title of chart
plt.title('Death by Race')
  
# Add Legends
plt.legend(labels, loc="upper right")
  
# Displaing Chart
plt.show()

Create a Chart Comparing the Total Number of Deaths of Men and Women

Use df_fatalities to illustrate how many more men are killed compared to women.

df_fatalities["gender"].value_counts()

M    2428
F     107
Name: gender, dtype: int64

Create a Box Plot Showing the Age and Manner of Death

Break out the data by gender using df_fatalities. Is there a difference between men and women in the manner of death?

df_fatalities_female=df_fatalities[df_fatalities["gender"]=="F"]

df_fatalities_male=df_fatalities[df_fatalities["gender"]=="M"]

fig = plt.figure(figsize =(10, 7))
ax = fig.add_axes([0, 0, 1, 1])
# Creating plot
data=[df_fatalities_female["age"],df_fatalities_male["age"]]
bp = ax.boxplot(data)
plt.show()

df_fatalities_female_fg=df_fatalities_female.groupby("manner_of_death")
df_fatalities_female_fg=pd.DataFrame(df_fatalities_female_fg["armed"].count()).reset_index().rename({'armed':'Count'})
df_fatalities_female_fg

	manner_of_death	armed
0	shot	102
1	shot and Tasered	5

df_fatalities_male_fg=df_fatalities_male.groupby("manner_of_death")
df_fatalities_male_fg=pd.DataFrame(df_fatalities_male_fg["armed"].count()).reset_index().rename({'armed':'Count'})
df_fatalities_male_fg

	manner_of_death	armed
0	shot	2261
1	shot and Tasered	167

Were People Armed?

In what percentage of police killings were people armed? Create chart that show what kind of weapon (if any) the deceased was carrying. How many of the people killed by police were armed with guns versus unarmed?

(df_fatalities[df_fatalities["armed"]=="unarmed"].count()/df_fatalities.count())*100

id                        6.75
name                      6.75
date                      6.75
manner_of_death           6.75
armed                     6.75
age                       6.75
gender                    6.75
race                      6.75
city                      6.75
state                     6.75
signs_of_mental_illness   6.75
threat_level              6.75
flee                      6.75
body_camera               6.75
dtype: float64

6.75% unarmed

df_fatalities_armed=df_fatalities.groupby("armed")
df_fatalities_armed=pd.DataFrame(df_fatalities_armed["city"].count()).reset_index()
df_fatalities_armed

	armed	city
0	0	9
1	Taser	9
2	air conditioner	1
3	ax	9
4	baseball bat	8
...	...	...
64	toy weapon	104
65	unarmed	171
66	undetermined	117
67	unknown weapon	18
68	vehicle	177

69 rows × 2 columns

df_fatalities_armed=df_fatalities_armed[1:]

fig = plt.figure(figsize = (12, 6))
 

plt.bar(df_fatalities_armed["armed"], df_fatalities_armed["city"], color ='maroon',
        width = 0.4)
 
plt.xlabel("Weapon")
plt.ylabel("Count")
plt.title("Weapon Carried by People")
plt.tick_params(axis='x', which='major', labelsize=9,rotation=90)
plt.show()

How Old Were the People Killed?

Work out what percentage of people killed were under 25 years old.

(df_fatalities[df_fatalities["age"]<25].count()/df_fatalities.count())*100

id                        20.79
name                      20.79
date                      20.79
manner_of_death           20.79
armed                     20.79
age                       20.79
gender                    20.79
race                      20.79
city                      20.79
state                     20.79
signs_of_mental_illness   20.79
threat_level              20.79
flee                      20.79
body_camera               20.79
dtype: float64

Create a histogram and KDE plot that shows the distribution of ages of the people killed by police.

sns.displot(df_fatalities, x="age", kind="kde")

<seaborn.axisgrid.FacetGrid at 0x16f2d6ed750>

df_fatalities_by_race=df_fatalities.groupby("race")

#df_fatalities_by_race=pd.DataFrame(df_fatalities_by_race)

df_fatalities_by_race.get_group("A")["age"]

0      53.00
69     35.00
155    35.00
160    28.00
265    28.00
303    39.00
337    32.00
468    60.00
685    59.00
686    31.00
766    44.00
899    25.00
945    39.00
987     0.00
1053   40.00
1203   38.00
1204   53.00
1293   41.00
1312   34.00
1322   28.00
1347   33.00
1480   27.00
1481   61.00
1560   18.00
1747   44.00
1790   45.00
1845   15.00
1883   26.00
1937   48.00
1991   37.00
1996   32.00
2052   41.00
2088   33.00
2198   45.00
2223   30.00
2235   33.00
2263   43.00
2367   18.00
2403   20.00
Name: age, dtype: float64

df_fatalities_dict=list(df_fatalities_by_race.groups.keys())
df_fatalities_dict=df_fatalities_dict[1:]
df_fatalities_dict

['A', 'B', 'H', 'N', 'O', 'W']

Create a seperate KDE plot for each race. Is there a difference between the distributions?

for i in range(len(df_fatalities_dict)):
    sns.displot(df_fatalities_by_race.get_group(df_fatalities_dict[i]), x="age", kind="kde")
    plt.title(df_fatalities_dict[i])
    plt.show()
#in black younger people had more fatalities for Americans 30-60

Race of People Killed

Create a chart that shows the total number of people killed by race.

import matplotlib.pyplot as plt
  
 
# given values
sizes = values.values()
  
# Setting labels for items in Chart
labels = values.keys()

  
# explosion
explode = (0.05, 0.05, 0.05, 0.05, 0.05)
  
# Pie Chart
plt.pie(sizes, labels=labels,
        autopct='%1.1f%%', pctdistance=0.85)
  
# draw circle
centre_circle = plt.Circle((0, 0), 0.70, fc='white')
fig = plt.gcf()
  
# Adding Circle in Pie chart
fig.gca().add_artist(centre_circle)
  
# Adding Title of chart
plt.title('Death by Race')
  
# Add Legends
plt.legend(labels, loc="upper right")
  
# Displaing Chart
plt.show()

df_fatalities

	id	name	date	manner_of_death	armed	age	gender	race	city	state	signs_of_mental_illness	threat_level	flee	body_camera
0	3	Tim Elliot	02/01/15	shot	gun	53.00	M	A	Shelton	WA	True	attack	Not fleeing	False
1	4	Lewis Lee Lembke	02/01/15	shot	gun	47.00	M	W	Aloha	OR	False	attack	Not fleeing	False
2	5	John Paul Quintero	03/01/15	shot and Tasered	unarmed	23.00	M	H	Wichita	KS	False	other	Not fleeing	False
3	8	Matthew Hoffman	04/01/15	shot	toy weapon	32.00	M	W	San Francisco	CA	True	attack	Not fleeing	False
4	9	Michael Rodriguez	04/01/15	shot	nail gun	39.00	M	H	Evans	CO	False	attack	Not fleeing	False
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
2530	2822	Rodney E. Jacobs	28/07/17	shot	gun	31.00	M	0	Kansas City	MO	False	attack	Not fleeing	False
2531	2813	TK TK	28/07/17	shot	vehicle	0.00	M	0	Albuquerque	NM	False	attack	Car	False
2532	2818	Dennis W. Robinson	29/07/17	shot	gun	48.00	M	0	Melba	ID	False	attack	Car	False
2533	2817	Isaiah Tucker	31/07/17	shot	vehicle	28.00	M	B	Oshkosh	WI	False	attack	Car	True
2534	2815	Dwayne Jeune	31/07/17	shot	knife	32.00	M	B	Brooklyn	NY	True	attack	Not fleeing	False

2535 rows × 14 columns

Mental Illness and Police Killings

What percentage of people killed by police have been diagnosed with a mental illness?

(df_fatalities[df_fatalities["signs_of_mental_illness"]==True].count()/df_fatalities.count())*100
#24.97%

id                        24.97
name                      24.97
date                      24.97
manner_of_death           24.97
armed                     24.97
age                       24.97
gender                    24.97
race                      24.97
city                      24.97
state                     24.97
signs_of_mental_illness   24.97
threat_level              24.97
flee                      24.97
body_camera               24.97
dtype: float64

df_fatalities

	id	name	date	manner_of_death	armed	age	gender	race	city	state	signs_of_mental_illness	threat_level	flee	body_camera
0	3	Tim Elliot	02/01/15	shot	gun	53.00	M	A	Shelton	WA	True	attack	Not fleeing	False
1	4	Lewis Lee Lembke	02/01/15	shot	gun	47.00	M	W	Aloha	OR	False	attack	Not fleeing	False
2	5	John Paul Quintero	03/01/15	shot and Tasered	unarmed	23.00	M	H	Wichita	KS	False	other	Not fleeing	False
3	8	Matthew Hoffman	04/01/15	shot	toy weapon	32.00	M	W	San Francisco	CA	True	attack	Not fleeing	False
4	9	Michael Rodriguez	04/01/15	shot	nail gun	39.00	M	H	Evans	CO	False	attack	Not fleeing	False
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
2530	2822	Rodney E. Jacobs	28/07/17	shot	gun	31.00	M	0	Kansas City	MO	False	attack	Not fleeing	False
2531	2813	TK TK	28/07/17	shot	vehicle	0.00	M	0	Albuquerque	NM	False	attack	Car	False
2532	2818	Dennis W. Robinson	29/07/17	shot	gun	48.00	M	0	Melba	ID	False	attack	Car	False
2533	2817	Isaiah Tucker	31/07/17	shot	vehicle	28.00	M	B	Oshkosh	WI	False	attack	Car	True
2534	2815	Dwayne Jeune	31/07/17	shot	knife	32.00	M	B	Brooklyn	NY	True	attack	Not fleeing	False

2535 rows × 14 columns

In Which Cities Do the Most Police Killings Take Place?

Create a chart ranking the top 10 cities with the most police killings. Which cities are the most dangerous?

df_fatalities_cities=df_fatalities["city"].value_counts().head(10)

df_fatalities_cities
#Los Angeles is the most dangerous

Los Angeles    39
Phoenix        31
Houston        27
Chicago        25
Las Vegas      21
San Antonio    20
Columbus       19
Austin         18
Miami          18
St. Louis      15
Name: city, dtype: int64

Rate of Death by Race

Find the share of each race in the top 10 cities. Contrast this with the top 10 cities of police killings to work out the rate at which people are killed by race for each city.

df_fatalities_race_city=df_fatalities["race"].groupby(df_fatalities["city"]).value_counts().nlargest(10)
df_fatalities_race_city

city         race
Chicago      B       21
Los Angeles  H       19
Houston      B       15
Austin       W       13
Phoenix      W       12
San Antonio  H       12
Columbus     B       11
Phoenix      H       11
St. Louis    B       11
Los Angeles  B       10
Name: race, dtype: int64

df_fatalities_by_state=df_fatalities.groupby("state")
df_fatalities_by_state=pd.DataFrame(df_fatalities_by_state["armed"].count()).reset_index().rename({'armed':'Count'})
df_fatalities_by_state

	state	armed
0	AK	15
1	AL	50
2	AR	26
3	AZ	118
4	CA	424
5	CO	74
6	CT	9
7	DC	11
8	DE	8
9	FL	154
10	GA	70
11	HI	11
12	IA	12
13	ID	17
14	IL	62
15	IN	43
16	KS	24
17	KY	43
18	LA	57
19	MA	22
20	MD	38
21	ME	13
22	MI	37
23	MN	32
24	MO	64
25	MS	23
26	MT	11
27	NC	69
28	ND	4
29	NE	15
30	NH	7
31	NJ	35
32	NM	51
33	NV	42
34	NY	45
35	OH	79
36	OK	78
37	OR	38
38	PA	51
39	RI	2
40	SC	44
41	SD	10
42	TN	59
43	TX	225
44	UT	23
45	VA	47
46	VT	3
47	WA	62
48	WI	43
49	WV	27
50	WY	8

import plotly.express as px

fig = px.choropleth(locations=df_fatalities_by_state["state"], locationmode="USA-states", color=df_fatalities_by_state["armed"], scope="usa")
fig.show()

df_fatalities

	id	name	date	manner_of_death	armed	age	gender	race	city	state	signs_of_mental_illness	threat_level	flee	body_camera
0	3	Tim Elliot	02/01/15	shot	gun	53.00	M	A	Shelton	WA	True	attack	Not fleeing	False
1	4	Lewis Lee Lembke	02/01/15	shot	gun	47.00	M	W	Aloha	OR	False	attack	Not fleeing	False
2	5	John Paul Quintero	03/01/15	shot and Tasered	unarmed	23.00	M	H	Wichita	KS	False	other	Not fleeing	False
3	8	Matthew Hoffman	04/01/15	shot	toy weapon	32.00	M	W	San Francisco	CA	True	attack	Not fleeing	False
4	9	Michael Rodriguez	04/01/15	shot	nail gun	39.00	M	H	Evans	CO	False	attack	Not fleeing	False
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
2530	2822	Rodney E. Jacobs	28/07/17	shot	gun	31.00	M	0	Kansas City	MO	False	attack	Not fleeing	False
2531	2813	TK TK	28/07/17	shot	vehicle	0.00	M	0	Albuquerque	NM	False	attack	Car	False
2532	2818	Dennis W. Robinson	29/07/17	shot	gun	48.00	M	0	Melba	ID	False	attack	Car	False
2533	2817	Isaiah Tucker	31/07/17	shot	vehicle	28.00	M	B	Oshkosh	WI	False	attack	Car	True
2534	2815	Dwayne Jeune	31/07/17	shot	knife	32.00	M	B	Brooklyn	NY	True	attack	Not fleeing	False

2535 rows × 14 columns

Number of Police Killings Over Time

Analyse the Number of Police Killings over Time. Is there a trend in the data?

df_fatalities_by_year=df_fatalities.groupby(["date"])
df_fatalities_by_year=pd.DataFrame(df_fatalities_by_year["armed"].count()).reset_index().rename({'armed':'Count'})
df_fatalities_by_year

	date	armed
0	01/01/16	1
1	01/01/17	6
2	01/02/16	2
3	01/02/17	2
4	01/03/15	3
...	...	...
874	31/08/16	2
875	31/10/15	2
876	31/10/16	3
877	31/12/15	1
878	31/12/16	3

879 rows × 2 columns

df_fatalities_by_year['date']=df_fatalities_by_year['date'].dt.strftime('%d/%m/%Y')
sns.displot(df_fatalities_by_year, x="date", kind="kde")

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
C:\Users\SHRUTI~1\AppData\Local\Temp/ipykernel_3976/1666060661.py in <module>
----> 1 df_fatalities_by_year['date']=df_fatalities_by_year['date'].dt.strftime('%d/%m/%Y')
      2 sns.displot(df_fatalities_by_year, x="date", kind="kde")

~\AppData\Local\Programs\Python\Python310\lib\site-packages\pandas\core\generic.py in __getattr__(self, name)
   5485         ):
   5486             return self[name]
-> 5487         return object.__getattribute__(self, name)
   5488 
   5489     def __setattr__(self, name: str, value) -> None:

~\AppData\Local\Programs\Python\Python310\lib\site-packages\pandas\core\accessor.py in __get__(self, obj, cls)
    179             # we're accessing the attribute of the class, i.e., Dataset.geo
    180             return self._accessor
--> 181         accessor_obj = self._accessor(obj)
    182         # Replace the property with the accessor object. Inspired by:
    183         # https://www.pydanny.com/cached-property.html

~\AppData\Local\Programs\Python\Python310\lib\site-packages\pandas\core\indexes\accessors.py in __new__(cls, data)
    504             return PeriodProperties(data, orig)
    505 
--> 506         raise AttributeError("Can only use .dt accessor with datetimelike values")

AttributeError: Can only use .dt accessor with datetimelike values

Epilogue

Now that you have analysed the data yourself, read The Washington Post's analysis here.